Long-Running Batch Job Replaced Mid-Flight Without State Checkpoint

System Design Interview•~30 min read•medium

Nightly 6-hour financial calculation job running on a Kubernetes pod when an automated rolling deployment triggers.

1. Understand Requirements & Scope

Nightly 6-hour financial calculation job running on a Kubernetes pod when an automated rolling deployment triggers. Failure cascade dynamic: Scale Surge → Unbounded Contention → Physical Bottleneck → 💥 Outage Cascade.

Functional Requirements:

Operational Invariant: "Batch processing jobs must persist periodic checkpoint progress to enable resume-from-failure after pod terminations."
Production Hardening: Architect long-running batch jobs to be idempotent and checkpointed: save progress state periodically (e.g. after every 10,000 records) to database/S3, and handle SIGTERM gracefull
Prevent cascading failure by isolating downstream degradation at the ingress boundary.

2. Back-of-the-Envelope Capacity Estimation

• Incident Scenario Baseline: Nightly 6-hour financial calculation job running on a Kubernetes pod when an automated rolling deployment triggers.

• Failure Cascade Path: Scale Surge → Unbounded Contention → Physical Bottleneck → 💥 Outage Cascade

• Sub-millisecond rejection / shed latency for overloaded requests

3. High-Level Architecture Flow

Architect long-running batch jobs to be idempotent and checkpointed: save progress state periodically (e.g. after every 10,000 records) to database/S3, and handle SIGTERM gracefully by flushing the latest checkpoint before exit. Invariant: "Batch processing jobs must persist periodic checkpoint progress to enable resume-from-failure after pod terminations."

Ingress → Bounded Admission Controller → Adaptive Concurrency / Queue Gate → Protected Worker Engine; Overload Shed Path → HTTP 429 / Circuit Fallback (<0.1ms).

Fig 1. Architecture data flow and component topology.

4. Production Implementation & Code

// HARDENED: Fast local Bloom filter / Redis revocation list verification
Claims claims = Jwts.parser().setSigningKey(secret).parseClaimsJws(token).getBody();
if (revocationBloomFilter.mightContain(claims.getId()) && redis.exists("revoked:" + claims.getId())) {
    throw new SecurityException("Revoked token access attempt rejected");
}

Hardened Implementation Sequence

Ingress Boundary Defense: Architect long-running batch jobs to be idempotent and checkpointed: save progress state periodically (e.g. after every 10,000 records) to database/S3
Failure Containment: Eliminate unconstrained buffering and propagate physical backpressure.
Invariant Enforcement: Batch processing jobs must persist periodic checkpoint progress to enable resume-from-failure after pod terminations.

5. Key Decisions & Operational Mitigations

Architectural Decisions & Trade-Offs

•Ingress Admission: Naive unconstrained buffering causes queue bloat and GC exhaustion. Mitigation: Bounded ingress gate with fast non-blocking shedding.
•Resource Strategy: Static capacity allocations collapse under brownouts. Mitigation: Dynamic adaptive limits scaling to real-time physical bounds.
•Outage Impact: Cascading lock convoys and memory saturation. Mitigation: Isolated failure domains keep P99 latency flat and protected.

Operational Hazards & Mitigations

•Cascading Failure: Deployment terminates the pod at hour 5. The job aborts without saving intermediate state. The new pod restarts from hour 0, missing business morning deadlines and calculating duplicate intermediate balances.
•Breakdown Chain: Scale Surge → Unbounded Contention → Physical Bottleneck → 💥 Outage Cascade
•Shedding Failure: Without fast rejection at the boundary, queued dead work triggers multi-second GC pauses, CPU thrashing, and cascading 504 Gateway Timeouts.

Interactive Knowledge Check • Question 1 of 3↓

Interactive Practice Quiz•Question 1 of 3

intermediate

Canary vs Blue-Green Release Strategies

What is the key advantage of Canary Deployments over Blue-Green Deployments for large-scale microservice clusters?